SOCIAL PERCEPTION SEQUENTIAL RECOMMENDER SYSTEMS MODEL BASED ON GRAPH NEURAL NETWORK
Since user preferences are dynamic and changeable and are affected by social relationships,traditional recommendation methods are often incomplete.To solve this problem,a social perception sequential recommendation model based on graph neural network(GASR)is proposed.The dynamic interest extraction layer was designed to capture the user's dynamic preferences.The social perception layer was designed and the graph neural network was used to construct the user's social relationship graph.The attention aggregation method was used to weigh the influence of different friends on user preferences.Experimental results on two actual data sets show that the model is superior to the latest social recommendation models and several competitive baseline models.
万方数据
维普
